Imatinib disrupts lymphoma angiogenesis by targeting vascular pericytes

Abstract

Pericytes and vascular smooth muscle cells (VSMCs), which are recruited to developing blood vessels by platelet-derived growth factor BB, support endothelial cell survival and vascular stability. Here, we report that imatinib, a tyrosine kinase inhibitor of platelet-derived growth factor receptor β (PDGFRβ), impaired growth of lymphoma in both human xenograft and murine allograft models. Lymphoma cells themselves neither expressed PDGFRβ nor were growth inhibited by imatinib. Tumor growth inhibition was associated with decreased microvascular density and increased vascular leakage. In vivo, imatinib induced apoptosis of tumor-associated PDGFRβ(+) pericytes and loss of perivascular integrity. In vitro, imatinib inhibited PDGFRβ(+) VSMC proliferation and PDGF-BB signaling, whereas small interfering RNA knockdown of PDGFRβ in pericytes protected them against imatinib-mediated growth inhibition. Fluorescence-activated cell sorter analysis of tumor tissue revealed depletion of pericytes, endothelial cells, and their progenitors following imatinib treatment. Compared with imatinib, treatment with an anti-PDGFRβ monoclonal antibody partially inhibited lymphoma growth. Last, microarray analysis (Gene Expression Omnibus database accession number GSE30752) of PDGFRβ(+) VSMCs following imatinib treatment showed down-regulation of genes implicated in vascular cell proliferation, survival, and assembly, including those representing multiple pathways downstream of PDGFRβ. Taken together, these data indicate that PDGFRβ(+) pericytes may represent a novel, nonendothelial, antiangiogenic target for lymphoma therapy.

Figures

Figure 1

Imatinib impaired growth of human lymphoma xenografts by disruption of tumor-associated microvasculature. (A-C) Tumor growth curves based on tumor volumes (mm3), comparing imatinib mesylate treatment (red) vs PBS control (black). (D-F) Tumor weight (g) at the time of tissue harvest, comparing imatinib (red) vs PBS (black). (G-I) Microvessels delineated by CD31 stain (red) in (G) PBS- and (H) imatinib-treated tumors, and (I) microvessel density normalized to PBS control. (J-L) Confocal DIC capture of functional vascular flow measured by isolectin (red) and normalized to PBS control. (M-P) Confocal analysis of pericyte marker staining in tumors. Blue marks functional vascular flow measured by isolectin, red outlines PDGFRβ+ pericytes, and green shows cleaved caspase 3 in apoptotic cells. (M) PBS-treated tumor. (N) Imatinib-treated tumor, which displays microvascular disintegration and flow leakage (white arrows), and patchy areas of tumor apoptosis/necrosis. (O) White arrows indicate apoptotic, yet structurally intact, PDGFRβ+ cells in close proximity to functional flow. (P) White arrows mark apoptosis of isolectin+ endothelial cells, and white asterisk (*) indicates leakage of infused intravascular isolectin. Scale bar, 50 μm.

Figure 2

Imatinib treatment of human DLBCL tumors was associated with pericytic dropout and disruption of microvasculature. (A1-A4) Analysis of α-SMA+ pericytes and CD31+ vessels in (A1,A3) Karpas422 and (A2,A4) OCI-Ly7 xenografts. (B1-B4) Analysis of NG2+ pericytes and CD31+ vessels in (B1,B3) Karpas422 and (B2,B4) OCI-Ly7 xenografts. (C) Confocal analysis of pericytes (red) in relation to intraluminal blood flow (isolectin, blue) in OCI-Ly7 tumors. (C1-C2) Two z-stack images focused at different depths within the same tissue; white arrows indicate disrupted microvessels with scanty PDGFRβ+ staining and abundant cleaved-caspase 3+, apoptotic nuclei; white asterisks (*) indicate microvessels with relatively intact blood flow and perivascular PDGFRβ+ staining. (C3-C4) White arrows indicate apoptotic α-SMA+ pericytes surrounding regions of functional blood flow. (D) Pericyte coverage was quantified as α-SMA+ area (red) and normalized to the PBS control. (E) Pericyte coverage was quantified as NG2+ area (green) and normalized to the PBS control. (F) Microvessel density was quantified as CD31+ staining area (blue) and normalized to the PBS control. *P < .05 compared with control. Scale bar, 50 μm.

Figure 3

Imatinib treatment of murine EL4 tumors impaired growth, decreased pericyte coverage, and reduced vascularization. (A) Tumor weight (g) at the time of tissue harvest, comparing imatinib (red) vs PBS (black). (B) Analysis of pericytes with (1-2) NG2, (3-4) PDGFRβ, and (5-6) α-SMA staining, and CD31+ vessels in EL4 tumors, in response to either imatinib or PBS. (C) Microvessel density was quantified as CD31+ staining density, and normalized to the PBS control. Pericyte coverage was quantified as NG2+, PDGFRβ+, or α-SMA+ area and normalized to the PBS control. *P < .05. Scale bar, 50 μm.

Figure 4

FACS analysis of lymphoma vascular cell populations. (A) Quad-gate analysis of the EL4 tumor-associated cells. Murine EPCs (mCD45−VEGFR2+CD117+), mature ECs (mCD45−VEGFR2+CD31+), murine PPCs (mCD45−CD140b+CD117+), and mature PCs (mCD45−CD140b+NG2+) are shown. Blue-dotted squares outline individual quad-gates for EPCs, PPCs, ECs, and PCs. (B) Quantification of stromal cell populations in EL4 tumors treated with imatinib vs PBS. *P < .05 compared with control.

Figure 5

Imatinib inhibited PDGFRβ signaling in vascular mural cells. (A) FACS analysis of PDGFRβ expression in VSMC and lymphoma cells. (B) Immunoblot analysis of PDGFRβ protein expression in stromal and lymphoma cells. (C) Imatinib-mediated growth inhibition in VSMC, endothelial cells, and lymphoma cells. (D) Relative abundance of PDGFRβ mRNA at 24, 48, and 72 hours following transfection of siRNA #1 targeting PDGFRβ in HBVPs, and expressed as the percentage normalized to expression level at 24 hours after transfection. (E) Viable cell counts at 48 hours after treatment with imatinib (20 μM for HBVP, 50 μM for Karpas422), following transfection with siRNA for 24 hours in HBVPs and Karpas422 cells. All cell counts were normalized to values at 24 hours after siRNA transfection. (F) Viable cell counts at 24, 48, and 72 hours after transfection of siRNA nontargeting (NT) or PDGFRβ-directed siRNA #1 in HBVPs, Karpas422, and Farage DLBCL cell lines, and normalized to the values at 24 hours after transfection. Data are representative of triplicate experiments in C-F. (G) Immunostaining for phospho-PDGFRβ in NG2+ pericytes in Karpas422 tumors treated with either (1) PBS or (2) imatinib. White arrow indicates expression of phospho-PDGFRβ in NG2+ pericytes. (3) PDGFRβ signaling was quantified as phospho-PDGFRβ+ area and normalized to PBS control. *P < .05 compared with control. Scale bar, 50 μm.

Figure 6

Imatinib blocked angiogenic pathways in vascular mural cells. (A) Effect of imatinib on VEGF secretion from VSMCs at 3, 24, and 48 hours after treatment. (B) Effect of imatinib on TGF-β secretion from VSMCs and cardiac microvascular endothelial cells (CMECs) at 48 hours after treatment. (C) Effect of imatinib on PDGF-BB–induced phosphorylation of PDGFRα, PDGFRβ, c-Kit, c-Abl, AKT, and ERK1/2 by immunoblot of VSMCs and pericytes in culture.

Figure 7

Anti-PDGFRβ antibody 2C5 inhibited growth of Farage xenografts in SCID mice. (A) Tumor growth curves based on tumor volumes (mm3) for PBS or hIgG control (black squares), imatinib (red circles), and 2C5 (blue triangles) treatments. (B) Tumor weight (g) at the time of tissue harvest comparing 2C5 (blue) vs imatinib (red) vs control (black). Inset depicts gross morphology of the resected tumor xenografts treated with (1) control, (2) imatinib, and (3) 2C5. (C) Staining of endothelial cells (CD31, red) and pericytes (NG2, green) in Farage xenografts treated with (1) control, (2) imatinib, and (3) 2C5. (4) NG2+ pericytic density and (5) CD31+ endothelial density were normalized to their respective controls. (D) Confocal analysis of pericytes in Farage tumors treated with (1) control, (2) imatinib, and (3) 2C5. Blue marks functional vascular flow by isolectin staining; red outlines NG2+ pericytes; and green stains CD68+ myelomonocytic cells. White arrows indicate perivascular CD68+ cells in close association with microvessels. *P < .05 compared with control. Scale bar, 50 μm.